unfunfunt / var_noise01 Goto Github PK

var_noise01
Mathew Galloway
2017

This project is a variable noise model generator for the SPIDER experiment. It takes the raw data and computes the power spectrum of each channel in a bunch of segments defined by the flagging. Each power spectrum is composed of a high frequency component that is created by averaging togeather 1024 point FFTs, and a low frequency component generated by using the Lomb Scargle periodigram. This allows us to use only the good data in the power estimation but still retain information about scales larger than the gaps in the data. 

Once generated, the power spectra are fit with a four parameter model per chunk, and then these parameters are saved to disk. They are then interpreted by python code that I also wrote that is a part of the larger SPIDER codebase, and is used to generate simulated noise timestreams with the proper frequency characteristics as a function of time.

The code is written in C, with python being used to interface with the spider_tools repository. The main loop can be found in noisepowerspec.c, with the other files containing various helper functions. C was chosen for speed, as running the code on the entire SPIDER flight takes up to 12 hours per focal plane. All the sim generation code is written in python, and can be found in unifile_sim.py in make_variable_noise_real() if you have access to the repository.